[Xpath-ng] Thoughts on work products
bob mcwhirter
bob at werken.com
Thu Nov 21 09:35:54 MST 2002
Howdy all--
Been watching the threads for a day or so and finally got some time to
chip in.
First, over in jaxen-land, I personally have no desire to care at all
about w3c-xpath-2.0. The fact we support (at this point) 4 different
InfoSet object models makes the imperfections and difficulties of
w3c-xpath-2.0 that much larger.
I definitely would like to track the xpath-ng thoughts and work them
into jaxen if they appear to be more enlightened than the w3c's effort.
Just off the top of my head, the one feature I'd like would be something
similar to the STL algorithms idioms. XPath 1.0 already gives us
collections and iteration, so applying some algorithm to the set or each
member would be good.
The fact that XPath 1.0 many times will consider a NodeSet to be only
the first node can hinder these types of operations.
Stronger definition of ordering would be cool, also. If there's any
guarantee of order, let's not call it a 'Set'. Also, let's not refer
to 'the first node in the set'. XPath 1.0 does include, implicitly,
several ideas regarding ordering, and so Jaxen has used a Lists for the
implementation. Seems odd to return a List when the spec says NodeSet.
I have no solution, but -something- to make working with namespaces
easier. Once a week we get mail from someone who wonders why their
XPath doesn't work, since they use the same ns-prefixes. I see the
value maintaining separate prefix->uri mappings for the XPath apart
from the documents to which they are applied, but it causes headaches.
Bare-minimum, default NS for xpath exprs would rock.
Anyhow, just rambling. Glad to be here.
-bob
More information about the Xpath-ng
mailing list